Wednesday, October 7, 2015

What do you mean, water structure?

Ah, water structure. What do we mean by it? How do we measure it? Elise Duboué-Dijon and Damien Laage revisit this old question with a close look at how various popular order parameters fare in describing the hydration shell of a hydrophobic solute in MD simulations (JPCB 119, 8406; 2015 – paper here). The tetrahedrality, local density, Voronoi cell shape and others are considered, and the correlations between them are in general not terribly strong: they are each tending to measure different things. But in any event, the perturbations around the small hydrophobic solute are rather small relative to the bulk: there is nothing iceberg-like here, nor is there any sign of significant heterogeneity. I think it would be fair to say that, rather than implying that water structure is best defined as “X”, we should conclude that “water structure” is an ill-defined concept. The authors also conclude that angular distortions offer the best measure of fluctuations in water reorientation dynamics.

I sense a meaty story in this one. Nascent membrane proteins emerging from the ribosome are assembled and integrated into the cell membrane with the aid of the translocon, a channel-like complex of proteins within the membrane. This complex has an hourglass shape and is filled with water, and the insertion of membrane proteins here has been considered as a simple process of hydrophobic partitioning. But it’s not so simple, according to Stephen White at the University of California at Irvine and colleagues (S. Capponi et al., PNAS 112, 9016; 2015 – paper here). They have performed MD simulations of the bacterial SecY translocon complex, and find that the water inside is very different from the bulk phase, having retarded rotational dynamics and aligned dipoles: in other words, it is decidedly “anomalous water”, suggesting that the translocon can’t simply be regarded as a protein-conducting pore. So any hydrophobic partitioning is likely to be more subtle than has been supposed, and we need to consider some degree of functional modification of the water properties: as the authors put it, “what is the partitioning free energy of solutes between water in bulk and water in restraining confined spaces?”

“If life can be considered as a massive self-assembly process, water seems to be a major driving force behind it.” There’s a nice way to begin a paper, and it’s how Vrushali Hande and Suman Chakrabarty of the CSIR National Chemical Laboratory in Pune start their simulation study of water ordering around hydrophobic polymers (JPCB 119, 11346; 2015 – paper here). They investigate specifically the notion introduced by Chandler and coworkers of a qualitative change in hydration at a length scale of around 1 nm. This depends, the authors say, on the conformation of the polymer. When it is extended, the tetrahedral ordering of the hydration shell is more or less insensitive to polymer chain length, because of the sub-nanometre scale of hydrogen bonding around the polymer chain. But in a collapsed conformation it’s a different story, with the hydration waters then dynamically coupled to fluctuations of the polymer. All the same, tetrahedral ordering doesn’t provide a strong signature of any order-disorder transition in the hydration layer, at least until chain lengths of around C40. But the authors say that this collapse itself is linked to fluctuations in the solvent in the manner discussed by Chandler et al., which can induce local dewetting.


The open and collapsed states of hydrophobic polymers in water, studied by Hande and Chakrabarty.

What is the state of water close to hydrophilic surfaces? There have been several experimental suggestions that this “interfacial water” has, over a nanoscale thickness, a viscosity several orders of magnitude greater than the bulk (e.g. Jinesh et al., Phys. Rev. Lett. 96, 166103; 2006). Andrei Sommer at Ulm and colleagues have recently argued that this interfacial water can be modified by irradiation with near-IR laser light (A. P. Sommer et al., Sci Rep. 5, 12029; 2015 – paper here). They now suggest that the gradient in viscosity that this would imply might explain why and how the rate of ATP synthesis changes in response to both reactive oxygen species and such irradiation. If ROS increase the hydrophilicity of the membrane in which the ATP synthase is embedded, they say, then this will increase the viscosity further and degrade the efficiency of this rotary device. By the same token, IR light decreases the viscosity and has a contrary effect on ATP synthesis. Note that the argument is only indirectly supported by the experiments described here, which are concerned only with measuring changes in the nanoindentation force for a diamond tip penetrating a water-coated hydrophilic metal surface due to laser irradiation, and interpreting them in terms of viscosity changes in the water film.

The controversy around water’s putative liquid-liquid phase continues. There have already been responses to David Limmer and David Chandler’s suggestion that the metastable LL transition reported in previous theoretical work is just an unequilibrated state that would eventually convert to ice (JCP 135, 134503; 2011 and 138, 214504; 2013). But now in a preprint, Frank Smallenburg and Francesco Sciortino say that, by modifying the bond flexibility of ST2 water, they can continuously tune the LL critical point until it moves into a regime where the liquid is more stable than ice – thereby, they say, negating any kinetic arguments for why this critical point is a phantom of the simulation technique

The degree of covalency of the hydrogen bond in water has been much debated. Thomas Kuhne at Paderborn and colleagues propose that this can be quantified by measuring components of the magnetic shielding tensor of the water hydrogens in NMR (H. Elgabarty et al., Nat. Commun. 6, 8318; 2015 – paper here). They define covalency as the amount of electron density transferred between hydrogen-bonded molecules and the associated stabilization energy, which they calculate in ab initio simulations to be, respectively, around 10 milli-electrons and 15 kJ/mol. They describe a calibration of the relationship between these quantities and the hydrogen magnetic shielding tensor that would enable their experimental determination.

The exchange of amide hydrogens in proteins with water can be used as a measure of protein structuring, flexibility, dynamics, and solvent exposure. But the mechanism by which it happens hasn’t been clear. Filip Persson and Bertil Halle show how even deeply buried parts of the polypeptide chain may become briefly exposed to water by conformational fluctuations (PNAS 112, 10383; 2015 – paper here). Their simulations of the bovine pancreatic trypsin inhibitor are long enough to identify the elusive “open” state by which proton exchange happens: a state, they propose, that requires the N-H hydrogen to be within 2.6 Å of at least two water molecules, and not involved in any intramolecular hydrogen-bonding. As well as the donor water molecule, the second water molecule is needed to solvate and stabilize the transient imidate ion formed after proton extraction from N-H, before it acquires a replacement proton from this second molecule. This “open” state exists for around 100 ps on average in all the amide groups studied here.

Bertil continues to probe conformational dynamics in an NMR study with Shuji Kaieda of water displacement within the cavity of a lipid binding protein (JPCB 119, 7957; 2015 – paper here). Conformational changes act to gate this water release, with fluctuations in a critical part of the protein determining the rate of passage of some highly ordered internal waters. The latter fall into three dynamical classes, with distinct survival times of the order of 1 ns (most of the waters are of this type), 100 ns and 6 μs.

Functionally relevant conformational fluctuations are also studied in a preprint sent to me by Tomotaka Oroguchi and Masayoshi Nakasako of Keio University. Their MD simulations suggest that the functional motions of an enzyme (glutamate dehydrogenase) are dominated by nanoscale wetting/drying transitions of a small number of hydration water molecules in a hydrophobic pocket (HS1) of the active site, along with stepwise association and dissociation of water clusters in a cylindrical hydrophilic crevice (HS2). The interpretation of behaviour at the hydrophobic site is supported by measurements of the catalytic rate of a mutant in which this hydrophobicity is lower. The combination of changing hydration states at the two sites makes the enzyme act as something of a hydraulic machine. This offers a nice illustration of how the vague idea of water-lubricated conformational flexibility in proteins can be united with more precise notions of nanoscale wetting and dehydration transitions.


Snapshots of different wetting states for the hydrpphobic pocket HS1 of glutamate dehydrogenase, along with a heat map relating solvent occupancy of this cleft (Q) to the separation of the “jaws” (d).


The “GDH machine”, driven by changes in hydration states in the hydrophobic (HS1) and hydrophilic (HS2) sites.

Ion channel selectivity is largely determined by electrostatic interactions with charged residues in the channel. But Vicente Aguilella and colleagues at the Universitat Jaume I in Castellón present calculations and simulations which challenge the idea that only solvent-accessible residues near the ion permeation pathway matter (E. García-Giménez et al. JPCB 119, 8475; 2015 – paper here). Looking in particular at bacterial porin OmpF, they say that many other charged residues, including buried ones, may affect the pore selectivity and that the dielectric properties of the protein therefore matter.

How does trehalose protect proteins from urea-induced denaturation? Subrata Paul and Sandip Paul at the Indian Institute of Technology in Assam explore that question via MD simulations of the hydration of the simple protein model N-methylacetamide (JPCB 119, 9820; 2015 – paper here). They find that trehalose displaces urea from the vicinity of the amide, and that amide-water hydrogen bonds are replaced by amide-trehalose H-bonds; thus trehalose will reduce the propensity of water to H-bond with a protein backbone, which would otherwise stabilize the denatured state. The results also largely support the notion that urea denaturation occurs via direct interactions rather than indirect effects on “water structure”.

Here is another introductory remark that encapsulates an issue rather splendidly: “If proteins had evolved to fold in a vacuum, thermodynamic experiments in the laboratory could have been straightforwardly interpreted by statistical energy landscape theory, just as model computer simulations with implicit solvent have been. Instead, the intimate involvement of the aqueous environment in the folding process made the uncovering of the principles of the energy landscape theory of protein folding a convoluted process.” This comes from a paper by Peyer Wolynes and colleagues (B. J. Sirovetz et al., JPCB 119, 11416; 2015 – paper here) in which a new model of protein folding (the associative memory, water mediated, structure and energy model, AWSEM) is used to map out the folding diagram for two proteins and explore hot, cold and pressure-induced denaturation. This model uses a coarse-grained force field that, among other things, captures water-mediated interactions. Using ubiquitin and λ-repressor as the test cases, the work shows that the model can supply a unified description of all of these cases that captures the key features of experimental measurements.


Representative structures of uniquitin in the native and denatured states in the AWSEM.

An interesting model system for studying water wires is described by Mihail Barboiu of the European Institute of Membranes in Montpellier (M. Barboiu et al., JPCB 119, 8707; 2015 – paper here). They look at self-assembled structures of a synthetic bola-amphiphile, which contain transverse pores that are hydrophilic, chiral and can contain helical water wires in which the waters are strongly orientationally ordered. Cations can permeate along these channels, offering a simple analogue of ion conduction through biomolecular water channels (for example, in gramicidin A). The authors say that selectivity of ion transport here is dominated by a subtle balance between the hydration and complexation energies of the ions.


Looking down the chiral water channels in crystals of a bola-amphiphile.

In somewhat related territory, Manish Kumar at Penn State University and colleagues describe a new class of artificial water channels that self-assemble into membrane-like structures (Y.-X. Shen et al., PNAS 112, 9810; 2015 – paper here). They call them peptide-appended pillar[5]arenes, which have a linked arene belt in the middle and short peptides extending above and below it, making a tubular structure. These molecules have been investigated before (summarized in Cragg & Sharma, Chem. Soc. Rev. 41, 597; 2012), but those described here are more hydrophobic and will insert at rather high concentration into lipid membranes, making the membrane water-permeable (3.5x10^8 water molecules per s, comparable to aquaporins). Simulations suggest that the channels seem to fluctuate between filled and empty states of water in a wetting/dewetting transition.


Pillar[5]arenes (A and B), and their insertion into a lipid membrane (C). D shows the water permeability.

Something quite different from my colleague John Hallsworth at Queen’s in Belfast and his coworkers, who ask “Is there a common water-activity limit for the three domains of life?” (A. Stevenson et al., Int. Soc. Microbial Ecol. J. 9, 1333; 2015 – paper here). They report that halophilic Archaea and Bacteria, and some xerophilic fungi, can all sustain viability at water activities as low as about 0.61.Could this point to a common physicochemical/thermodynamic origin for such a limiting value? If so, could there be astrobiological implications? (A good time to be thinking about that!)

Another unusual kind of contribution with prebiotic resonances comes from Atul Parikh of Nanyang Technological University in Singapore and coworkers, who report that giant vesicles filled with sugar solution and subjected to osmotic stress in a bath of lower sugar concentration may undergo damped cycles of expansion and contraction, accompanied by temporary rupture of the vesicle walls (K. Oglecka et al., eLife 3, e03695; 2014 – paper here). In the expanded phase, the vesicles are patchy, due to phase separation of cholesterol and phospholipids in the walls; in the contracted phase they are uniform. The researchers say that this might have offered a useful, even adaptive, response to the microenvironment for early protocells. There’s a nice phys.org story on the work here.